Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
22.12% |
25 / 113 |
|
20.00% |
1 / 5 |
CRAP | |
0.00% |
0 / 1 |
| Ref | |
22.12% |
25 / 113 |
|
20.00% |
1 / 5 |
648.09 | |
0.00% |
0 / 1 |
| sourceToDom | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
12 | |||
| processAttributeEmbeddedHTML | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| lintHandler | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
20 | |||
| domToWikitext | |
41.51% |
22 / 53 |
|
0.00% |
0 / 1 |
82.83 | |||
| diffHandler | |
0.00% |
0 / 32 |
|
0.00% |
0 / 1 |
90 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | // phpcs:disable MediaWiki.WhiteSpace.SpaceBeforeSingleLineComment.NewLineComment |
| 4 | |
| 5 | namespace Cite\Parsoid; |
| 6 | |
| 7 | use Closure; |
| 8 | use Exception; |
| 9 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 10 | use Wikimedia\Parsoid\DOM\Element; |
| 11 | use Wikimedia\Parsoid\Ext\DOMDataUtils; |
| 12 | use Wikimedia\Parsoid\Ext\DOMUtils; |
| 13 | use Wikimedia\Parsoid\Ext\ExtensionTagHandler; |
| 14 | use Wikimedia\Parsoid\Ext\ParsoidExtensionAPI; |
| 15 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 16 | |
| 17 | /** |
| 18 | * Simple token transform version of the Ref extension tag. |
| 19 | * @license GPL-2.0-or-later |
| 20 | */ |
| 21 | class Ref extends ExtensionTagHandler { |
| 22 | |
| 23 | /** @inheritDoc */ |
| 24 | public function sourceToDom( |
| 25 | ParsoidExtensionAPI $extApi, string $txt, array $extArgs |
| 26 | ): ?DocumentFragment { |
| 27 | // Drop nested refs entirely, unless we've explicitly allowed them |
| 28 | $parentExtTag = $extApi->parentExtTag(); |
| 29 | if ( $parentExtTag === 'ref' && empty( $extApi->parentExtTagOpts()['allowNestedRef'] ) ) { |
| 30 | return null; |
| 31 | } |
| 32 | |
| 33 | // The one supported case for nested refs is from the {{#tag:ref}} parser |
| 34 | // function. However, we're overly permissive here since we can't |
| 35 | // distinguish when that's nested in another template. |
| 36 | // The php preprocessor did our expansion. |
| 37 | $allowNestedRef = $extApi->inTemplate(); |
| 38 | |
| 39 | return $extApi->extTagToDOM( |
| 40 | $extArgs, |
| 41 | $txt, |
| 42 | [ |
| 43 | // NOTE: sup's content model requires it only contain phrasing |
| 44 | // content, not flow content. However, since we are building an |
| 45 | // in-memory DOM which is simply a tree data structure, we can |
| 46 | // nest flow content in a <sup> tag. |
| 47 | 'wrapperTag' => 'sup', |
| 48 | 'parseOpts' => [ |
| 49 | 'extTag' => 'ref', |
| 50 | 'extTagOpts' => [ 'allowNestedRef' => $allowNestedRef ], |
| 51 | // Ref content doesn't need p-wrapping or indent-pres. |
| 52 | // Treat this as inline-context content to get that b/c behavior. |
| 53 | 'context' => 'inline', |
| 54 | ], |
| 55 | ] |
| 56 | ); |
| 57 | } |
| 58 | |
| 59 | /** @inheritDoc */ |
| 60 | public function processAttributeEmbeddedHTML( |
| 61 | ParsoidExtensionAPI $extApi, Element $elt, Closure $proc |
| 62 | ): void { |
| 63 | $dataMw = DOMDataUtils::getDataMw( $elt ); |
| 64 | if ( isset( $dataMw->body->html ) ) { |
| 65 | $dataMw->body->html = $proc( $dataMw->body->html ); |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | /** @inheritDoc */ |
| 70 | public function lintHandler( |
| 71 | ParsoidExtensionAPI $extApi, Element $ref, callable $defaultHandler |
| 72 | ): bool { |
| 73 | $dataMw = DOMDataUtils::getDataMw( $ref ); |
| 74 | if ( isset( $dataMw->body->html ) ) { |
| 75 | $fragment = $extApi->htmlToDom( $dataMw->body->html ); |
| 76 | $defaultHandler( $fragment ); |
| 77 | } elseif ( isset( $dataMw->body->id ) ) { |
| 78 | $refNode = DOMCompat::getElementById( $extApi->getTopLevelDoc(), $dataMw->body->id ); |
| 79 | if ( $refNode ) { |
| 80 | $defaultHandler( $refNode ); |
| 81 | } |
| 82 | } |
| 83 | return true; |
| 84 | } |
| 85 | |
| 86 | /** @inheritDoc */ |
| 87 | public function domToWikitext( |
| 88 | ParsoidExtensionAPI $extApi, Element $node, bool $wrapperUnmodified |
| 89 | ) { |
| 90 | $startTagSrc = $extApi->extStartTagToWikitext( $node ); |
| 91 | $dataMw = DOMDataUtils::getDataMw( $node ); |
| 92 | if ( !isset( $dataMw->body ) ) { |
| 93 | return $startTagSrc; // We self-closed this already. |
| 94 | } |
| 95 | |
| 96 | $html2wtOpts = [ |
| 97 | 'extName' => $dataMw->name, |
| 98 | // FIXME: One-off PHP parser state leak. This needs a better solution. |
| 99 | 'inPHPBlock' => true |
| 100 | ]; |
| 101 | |
| 102 | if ( isset( $dataMw->body->html ) ) { |
| 103 | // First look for the extension's content in data-mw.body.html |
| 104 | $src = $extApi->htmlToWikitext( $html2wtOpts, $dataMw->body->html ); |
| 105 | } elseif ( isset( $dataMw->body->id ) ) { |
| 106 | // If the body isn't contained in data-mw.body.html, look if |
| 107 | // there's an element pointed to by body->id. |
| 108 | $bodyElt = DOMCompat::getElementById( $extApi->getTopLevelDoc(), $dataMw->body->id ); |
| 109 | |
| 110 | // So far, this is specified for Cite and relies on the "id" |
| 111 | // referring to an element in the top level dom, even though the |
| 112 | // <ref> itself may be in embedded content, |
| 113 | // https://www.mediawiki.org/wiki/Specs/HTML/Extensions/Cite#Ref_and_References |
| 114 | // FIXME: This doesn't work if the <references> section |
| 115 | // itself is in embedded content, since we aren't traversing |
| 116 | // in there. |
| 117 | |
| 118 | // If we couldn't find a body element, this is a bug. |
| 119 | // Add some extra debugging for the editing client (ex: VisualEditor) |
| 120 | if ( !$bodyElt ) { |
| 121 | $extraDebug = ''; |
| 122 | $firstA = DOMCompat::querySelector( $node, 'a[href]' ); |
| 123 | if ( $firstA ) { |
| 124 | $href = DOMCompat::getAttribute( $firstA, 'href' ); |
| 125 | if ( $href && str_starts_with( $href, '#' ) ) { |
| 126 | try { |
| 127 | $ref = DOMCompat::querySelector( $extApi->getTopLevelDoc(), $href ); |
| 128 | if ( $ref ) { |
| 129 | $extraDebug .= ' [doc: ' . DOMCompat::getOuterHTML( $ref ) . ']'; |
| 130 | } |
| 131 | } catch ( Exception $e ) { |
| 132 | // We are just providing VE with debugging info. |
| 133 | // So, ignore all exceptions / errors in this code. |
| 134 | } |
| 135 | if ( !$extraDebug ) { |
| 136 | $extraDebug = ' [reference ' . $href . ' not found]'; |
| 137 | } |
| 138 | } |
| 139 | } |
| 140 | $extApi->log( |
| 141 | 'error/' . $dataMw->name, |
| 142 | 'extension src id ' . $dataMw->body->id . ' points to non-existent element for:', |
| 143 | DOMCompat::getOuterHTML( $node ), |
| 144 | '. More debug info: ', |
| 145 | $extraDebug |
| 146 | ); |
| 147 | return ''; // Drop it! |
| 148 | } |
| 149 | |
| 150 | $hasRefName = strlen( $dataMw->attrs->name ?? '' ) > 0; |
| 151 | $hasFollow = strlen( $dataMw->attrs->follow ?? '' ) > 0; |
| 152 | |
| 153 | if ( $hasFollow ) { |
| 154 | $about = DOMCompat::getAttribute( $node, 'about' ); |
| 155 | $followNode = $about !== null ? DOMCompat::querySelector( |
| 156 | $bodyElt, "span[typeof~='mw:Cite/Follow'][about='{$about}']" |
| 157 | ) : null; |
| 158 | if ( $followNode ) { |
| 159 | $src = $extApi->domToWikitext( $html2wtOpts, $followNode, true ); |
| 160 | $src = ltrim( $src, ' ' ); |
| 161 | } else { |
| 162 | $src = ''; |
| 163 | } |
| 164 | } else { |
| 165 | if ( $hasRefName ) { |
| 166 | // Follow content may have been added as spans, so drop it |
| 167 | if ( DOMCompat::querySelector( $bodyElt, "span[typeof~='mw:Cite/Follow']" ) ) { |
| 168 | $bodyElt = DOMDataUtils::cloneNode( $bodyElt, true ); |
| 169 | foreach ( $bodyElt->childNodes as $child ) { |
| 170 | if ( DOMUtils::hasTypeOf( $child, 'mw:Cite/Follow' ) ) { |
| 171 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
| 172 | DOMCompat::remove( $child ); |
| 173 | } |
| 174 | } |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | $src = $extApi->domToWikitext( $html2wtOpts, $bodyElt, true ); |
| 179 | } |
| 180 | } else { |
| 181 | $extApi->log( 'error', 'Ref body unavailable for: ' . DOMCompat::getOuterHTML( $node ) ); |
| 182 | return ''; // Drop it! |
| 183 | } |
| 184 | |
| 185 | return $startTagSrc . $src . '</' . $dataMw->name . '>'; |
| 186 | } |
| 187 | |
| 188 | /** @inheritDoc */ |
| 189 | public function diffHandler( |
| 190 | ParsoidExtensionAPI $extApi, callable $domDiff, Element $origNode, |
| 191 | Element $editedNode |
| 192 | ): bool { |
| 193 | $origDataMw = DOMDataUtils::getDataMw( $origNode ); |
| 194 | $editedDataMw = DOMDataUtils::getDataMw( $editedNode ); |
| 195 | |
| 196 | if ( isset( $origDataMw->body->html ) && isset( $editedDataMw->body->html ) ) { |
| 197 | $origFragment = $extApi->htmlToDom( |
| 198 | $origDataMw->body->html, $origNode->ownerDocument, |
| 199 | [ 'markNew' => true ] |
| 200 | ); |
| 201 | $editedFragment = $extApi->htmlToDom( |
| 202 | $editedDataMw->body->html, $editedNode->ownerDocument, |
| 203 | [ 'markNew' => true ] |
| 204 | ); |
| 205 | return call_user_func( $domDiff, $origFragment, $editedFragment ); |
| 206 | } elseif ( isset( $origDataMw->body->id ) && isset( $editedDataMw->body->id ) ) { |
| 207 | $origId = $origDataMw->body->id; |
| 208 | $editedId = $editedDataMw->body->id; |
| 209 | |
| 210 | // So far, this is specified for Cite and relies on the "id" |
| 211 | // referring to an element in the top level dom, even though the |
| 212 | // <ref> itself may be in embedded content, |
| 213 | // https://www.mediawiki.org/wiki/Specs/HTML/Extensions/Cite#Ref_and_References |
| 214 | // FIXME: This doesn't work if the <references> section |
| 215 | // itself is in embedded content, since we aren't traversing |
| 216 | // in there. |
| 217 | $origHtml = DOMCompat::getElementById( $origNode->ownerDocument, $origId ); |
| 218 | $editedHtml = DOMCompat::getElementById( $editedNode->ownerDocument, $editedId ); |
| 219 | |
| 220 | if ( $origHtml && $editedHtml ) { |
| 221 | return call_user_func( $domDiff, $origHtml, $editedHtml ); |
| 222 | } else { |
| 223 | // Log error |
| 224 | if ( !$origHtml ) { |
| 225 | $extApi->log( |
| 226 | 'error/domdiff/orig/ref', |
| 227 | "extension src id {$origId} points to non-existent element for:", |
| 228 | DOMCompat::getOuterHTML( $origNode ) |
| 229 | ); |
| 230 | } |
| 231 | if ( !$editedHtml ) { |
| 232 | $extApi->log( |
| 233 | // use info level to avoid logspam for CX edits where translated |
| 234 | // docs might reference nodes not copied over from orig doc. |
| 235 | 'info/domdiff/edited/ref', |
| 236 | "extension src id {$editedId} points to non-existent element for:", |
| 237 | DOMCompat::getOuterHTML( $editedNode ) |
| 238 | ); |
| 239 | } |
| 240 | } |
| 241 | } |
| 242 | |
| 243 | // FIXME: Similar to DOMDiff::subtreeDiffers, maybe $editNode should |
| 244 | // be marked as inserted to avoid losing any edits, at the cost of |
| 245 | // more normalization |
| 246 | |
| 247 | return false; |
| 248 | } |
| 249 | } |